# Magic getter bug tickled by Moodle - Thread: https://externals.io/message/81467 - Original archive: https://news-web.php.net/php.internals/81467 - Started by Rasmus Lerdorf on 2015-01-31T05:04:30+00:00 - Messages: 5 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #81467 — Rasmus Lerdorf — 2015-01-31T05:04:30+00:00 URL: https://externals.io/message/81467#81467 There is something weird going on with PHP7 and the moodle_page class and all its magic methods. It dies with a: /var/www/moodle/blocks/navigation/block_navigation.php(223) : Notice - Undefined property: moodle_page::$navigation /var/www/moodle/blocks/navigation/block_navigation.php(223) : Fatal error - __clone method called on non-object That code is: 222 $this->page->navigation->initialise(); 223 return clone($this->page->navigation); See: https://github.com/moodle/moodle/blob/master/blocks/navigation/block_navigation.php#L222-L223 $this->page_navigation is created on that initial call through the wonders of this magic: https://github.com/moodle/moodle/blob/master/lib/pagelib.php#L757-L772 Which somehow leads us to read 8 bytes in a free'ed block: https://gist.github.com/anonymous/f9d1f53b1205d29d86cc but I am not having much luck dwindling it down to a simpler reproduce case. If you know this code or have a couple of hours to spare, could you see if you could reduce it to something more manageable for us? (and yes, same code is fine in 5.6) -Rasmus --- ## #81469 — Michael Wallner — 2015-01-31T07:13:16+00:00 URL: https://externals.io/message/81467#81469 · In reply to: #81467 > > There is something weird going on with PHP7 and the moodle_page class > and all its magic methods. It dies with a: > > /var/www/moodle/blocks/navigation/block_navigation.php(223) : Notice - > Undefined property: moodle_page::$navigation > /var/www/moodle/blocks/navigation/block_navigation.php(223) : Fatal > error - __clone method called on non-object > > That code is: > > 222 $this->page->navigation->initialise(); > 223 return clone($this->page->navigation); > > See: > https://github.com/moodle/moodle/blob/master/blocks/navigation/block_navigation.php#L222-L223 > > $this->page_navigation is created on that initial call through the > wonders of this magic: > > https://github.com/moodle/moodle/blob/master/lib/pagelib.php#L757-L772 > > Which somehow leads us to read 8 bytes in a free'ed block: > > https://gist.github.com/anonymous/f9d1f53b1205d29d86cc > > but I am not having much luck dwindling it down to a simpler reproduce > case. If you know this code or have a couple of hours to spare, could > you see if you could reduce it to something more manageable for us? > > (and yes, same code is fine in 5.6) > Here's a similar bug, too: https://bugs.php.net/bug.php?id=68933&edit=1 --- ## #81470 — Rasmus Lerdorf — 2015-01-31T07:15:33+00:00 URL: https://externals.io/message/81467#81470 · In reply to: #81469 > > On 31 Jan 2015 06:04, "Rasmus Lerdorf" > wrote: > > […] > https://github.com/moodle/moodle/blob/master/blocks/navigation/block_navigation.php#L222-L223 > > […] > > Here's a similar bug, too: https://bugs.php.net/bug.php?id=68933&edit=1 Ah, good catch. Yes, https://gist.github.com/arjenschol/3d94195ca51aa44db1c6 looks like a similar pattern to what Moodle is doing in that part of the code that seems to trigger this. -Rasmus --- ## #81477 — Xinchen Hui — 2015-01-31T10:15:29+00:00 URL: https://externals.io/message/81467#81477 · In reply to: #81470 Hey: > > > […] > > Ah, good catch. Yes, > https://gist.github.com/arjenschol/3d94195ca51aa44db1c6 looks like a > similar pattern to what Moodle is doing in that part of the code that > seems to trigger this. must be fixed, https://github.com/php/php-src/commit/1a60175e2595a24ebc3b6d80a112d574c6c98f58 thanks > > -Rasmus > -- Xinchen Hui @Laruence http://www.laruence.com/ --- ## #81566 — Yasuo Ohgaki — 2015-02-02T08:52:59+00:00 URL: https://externals.io/message/81467#81566 · In reply to: #81477 Hi Xinchen, > must be fixed, > > https://github.com/php/php-src/commit/1a60175e2595a24ebc3b6d80a112d574c6c98f58 > Impressive work! It's irrelevant probably. I only get one valgrind leak with TS build now. NTS is fine. [yohgaki@dev php-src]$ USE_ZEND_ALLOC=0 valgrind --track-origins=yes --leak-check=full ./sapi/cli/php -r "echo 1;" ==13614== Memcheck, a memory error detector ==13614== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al. ==13614== Using Valgrind-3.10.1 and LibVEX; rerun with -h for copyright info ==13614== Command: ./php-bin -r echo\ 1; ==13614== 1==13614== ==13614== HEAP SUMMARY: ==13614== in use at exit: 88,808 bytes in 2,768 blocks ==13614== total heap usage: 32,420 allocs, 29,652 frees, 3,656,230 bytes allocated ==13614== ==13614== 376 bytes in 1 blocks are definitely lost in loss record 89 of 135 ==13614== at 0x4C29BCF: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13614== by 0xAEB3B5: alloc_globals_ctor (zend_alloc.c:2382) ==13614== by 0xA6665C: ts_allocate_id (TSRM.c:259) ==13614== by 0xAEB4B9: start_memory_manager (zend_alloc.c:2405) ==13614== by 0xB21D09: zend_startup (zend.c:572) ==13614== by 0xA6DA0F: php_module_startup (main.c:2105) ==13614== by 0xBD2FB1: php_cli_startup (php_cli.c:421) ==13614== by 0xBD53B7: main (php_cli.c:1341) ==13614== ==13614== LEAK SUMMARY: ==13614== definitely lost: 376 bytes in 1 blocks ==13614== indirectly lost: 0 bytes in 0 blocks ==13614== possibly lost: 0 bytes in 0 blocks ==13614== still reachable: 88,432 bytes in 2,767 blocks ==13614== suppressed: 0 bytes in 0 blocks ==13614== Reachable blocks (those to which a pointer was found) are not shown. ==13614== To see them, rerun with: --leak-check=full --show-leak-kinds=all ==13614== ==13614== For counts of detected and suppressed errors, rerun with: -v ==13614== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 0 from 0) -- Yasuo Ohgaki yohgaki@ohgaki.net